Development of Air-Stable, Supported Membrane Arrays with

Anal. Chem. 2008, 80, 6397–6404

Development of Air-Stable, Supported Membrane Arrays with Photolithography for Study of Phosphoinositide-Protein Interactions Using Surface Plasmon Resonance Imaging Zhuangzhi Wang, Thomas Wilkop, Jong Ho Han, Yi Dong, Matthew J. Linman, and Quan Cheng* Department of Chemistry, University of California, Riverside, California 92521 We report the development of an air-stable, supported membrane array by use of photolithography for label-free detection of lipid-protein interactions. Phosphoinositides and their phosphorylated derivatives (PIPs) were studied for their binding properties to proteins with lipid microarray in combination with surface plasmon resonance imaging (SPRi). We have demonstrated a simple method to fabricate lipid arrays using photoresist and carried out a series of surface characterizations with SPRi, ac impedance, cyclic voltammetry, and fluorescence microscopy to validate the array quality and lipid bilayer formation. A number of lipid compositions have been tested for the robustness of resulting membranes when undergoing dehydration and rehydration procedures, and the 1,2dioleoyl-sn-glycero-3-ethylphosphocholine/poly(ethylene glycol)-phosphatidylethanolamine (DOPC+/PEG-PE) system stands out as the best performer that yields the recovery to within 2% of the original state according to SPR sensorgrams. Limits of detection on the dehydrated/ rehydrated DOPC+/PEG-PE membranes were determined to be 33 nM for avidin binding to biotinylated lipids, 73.5 nM for cholera toxin to GM1, and 25 nM for PtdIns(4,5)P2-binding protein (P4,5BP) to PtdIns(4,5)P2 lipid, respectively. These results demonstrate the suitability and sensitivity of this membrane for constructing membrane arrays for SPRi analysis under ambient conditions. With the use of this addressable and functional lipid membrane array, the screening of specific lipid-protein interactions has been conducted. Strong and specific interactions between P4,5BP and PtdIns(4,5)P2/DOPC+/ PEG-PE membrane were observed as expected, while cross reactions were spotted for P4,5BP/PtdIns(4)P and avidin/GM1 at varied degrees. The air-stable membrane array demonstrated here presents a simple, effective approach for constructing functional membrane surfaces for screening applications, which opens a new avenue for the label-free study of membrane proteins and other forms of lipid-membrane interactions. Lipidomics, the systems-level analysis of lipids and their interacting network, is a relatively young field in biomedical * Corresponding author. E-mail: [email protected]. Phone: (951) 827-2702. 10.1021/ac800845w CCC: $40.75  2008 American Chemical Society Published on Web 07/12/2008

research and yet has attracted considerable attention recently.1 Lipids are not only the building components of the cell membrane but also substrates in cellular metabolism. The degradation of lipid metabolism has been associated with a variety of diseases, such as cardiovascular disease, cancer, obesity, and diabetes.2–7 In fact, lipid-protein interaction constitutes one of the most important interactions in biology and mediate numerous signal transduction pathways in addition to regulating cellular functions.8 The study of lipid-protein interactions can provide many new biological targets for therapeutic intervention and diagnosis.9 However, current methods to examine lipid-protein interactions have significant limitations. Principal methods employed today for these studies include protein-lipid overlay, enzymelinked immunosorbent assay (ELISA), and polymeric beads,10 where protein-lipid overlay assay is the most commonly practiced. In this method, lipids of interest such as phosphoinositides are directly immobilized on the surface either covalently11 or noncovalently,12 allowed to interact with its binding protein, and detected with a labeled antibody. Although this method allows for screening of affinities in a semiquantitative manner, it neglects the complex cell matrix effect since the lipids are not in a membrane environment. This can severely compromise the validity of (1) Wenk, M. R. Nat. Rev. Drug Discovery 2005, 4, 594–610. (2) Laporte, J.; Hu, L. J.; Kretz, C.; Mandel, J. L.; Kioschis, P.; Coy, J. F.; Klauck, S. M.; Poustka, A.; Dahl, N. Nat. Genet. 1996, 13, 175–182. (3) Li, S. L.; Tiab, L.; Jiao, X. D.; Munier, F. L.; Zografos, L.; Frueh, B. E.; Sergeev, Y.; Smith, J.; Rubin, B.; Meallet, M. A.; Forster, R. K.; Hejtmancik, J. F.; Schorderet, D. F. Am. J. Hum. Genet. 2005, 77, 54–63. (4) Li, J.; Yen, C.; Liaw, D.; Podsypanina, K.; Bose, S.; Wang, S. I.; Puc, J.; Miliaresis, C.; Rodgers, L.; Mccombie, R.; Bigner, S. H.; Giovanella, B. C.; Ittmann, M.; Tycko, B.; Hibshoosh, H.; Wigler, M. H.; Parsons, R. Science 1997, 275, 1943–1947. (5) Bader, A. G.; Kang, S. Y.; Vogt, P. K. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 1475–1479. (6) Kang, S. Y.; Bader, A. G.; Vogt, P. K. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 802–807. (7) Shayesteh, L.; Lu, Y. L.; Kuo, W. L.; Baldocchi, R.; Godfrey, T.; Collins, C.; Pinkel, D.; Powell, B.; Mills, G. B.; Gray, J. W. Nat. Genet. 1999, 21, 99– 102. (8) Feng, L.; Ferguson, C. G.; Drees, B. E.; Neilsen, P. O.; Prestwich, G. D. In Chemical Genomics; Darvas, F., Guttman, A., Dorman, G., Eds.; Marcel Dekker Inc.: New York, 2004; pp 215-274. (9) Prestwich, G. D. Prostaglandins Other Lipid Mediators 2005, 77, 168–178. (10) Lam, A. D.; Tryoen-Toth, P.; Tsai, B.; Vitale, N.; Stuenkel, E. L. Mol. Biol. Cell 2008, 19, 485–497. (11) Prestwich, G. D. Chem. Biol. 2004, 11, 619–637. (12) Guillou, H.; Lecureuil, C.; Anderson, K. E.; Suire, S.; Ferguson, G. J.; Ellson, C. D.; Gray, A.; Divecha, N.; Hawkins, P. T.; Stephens, L. R. J. Lipid Res. 2007, 48, 726–732.

Analytical Chemistry, Vol. 80, No. 16, August 15, 2008

6397

analytical measurements with regards to their application toward actual cell membrane lipid-protein interactions. The liposome pull-down assay, a new method which incorporates lipid components in a membrane-like vesicle structure, has also been frequently employed. However, this assay is tedious and does not provide quantitative results.13 A broader problem is that these methods are usually low throughput, which has become a bottleneck for the global profiling of lipid-protein interactions. Clearly, there is a compelling need for new analytical approaches for a high-throughput study of lipid-protein interactions in a membrane-like environment. Recently, surface plasmon resonance (SPR) and surface plasmon resonance imaging (SPRi) have become methods of choice to quantitatively examine lipid-protein interactions.14 SPR is an optical technique that can sensitively detect changes of the refractive index of a thin film15,16 and shows two major advantages over other techniques for interaction analysis: nonlabeling and real-time measurement. In the past 2 decades, SPR spectroscopy has been advanced to play a vital role in biospecific interaction analysis (BIA).17–20 SPR imaging further introduces high-throughput capability that allows for performing parallel bioanalysis in a complex matrix in a single run.21 Since it was first described by Rothenhausler and Knoll,22 SPRi has been used extensively in the studies of DNA, RNA, protein, and small molecule interactions.23–27 Supported lipid membranes (SLMs) or lipid membranes arrays generated on a solid substrate coupled with SPR imaging offer a promising route to the development of high-throughput lipidomics studies. SLMs have received considerable interest as sensing interfaces in the past few years, and their applications in sensor arrays for cell adhesion, on-chip immunoassays, and detection of warfare agents have been reported.28 Our group in particular has established supported lipid membranes on polydimethylsiloxane (PDMS), glass, and gold surfaces for a variety of sensing applications.29–31 To extend our studies to high throughput analysis, a key issue to be addressed is to find an effective (13) Wysocka, J. Methods 2006, 40, 339–343. (14) Besenicar, M.; Macek, P.; Lakey, J. H.; Anderluh, G. Chem. Phys. Lipids 2006, 141, 169–178. (15) Raether, H. Springer Tracts Mod. Phys. 1988, 111, 1–133. (16) Kretschmann, E.; Raether, H. Z. Naturforsch. 1968, 23A, 2135–2136. (17) Yoshida, T.; Sato, M.; Ozawa, T.; Umezawa, Y. Anal. Chem. 2000, 72, 6– 11. (18) Wink, T.; Van Zuilen, S. J.; Bult, A.; Van Bennekom, W. P. Anal. Chem. 1998, 70, 827–832. (19) Rich, R. L.; Day, Y. S. N.; Morton, T. A.; Myszka, D. G. Anal. Biochem. 2001, 296, 197–207. (20) Rich, R. L.; Myszka, D. G. Curr. Opin. Biotechnol. 2000, 11, 54–61. (21) Brockman, J. M.; Nelson, B. P.; Corn, R. M. Annu. Rev. Phys. Chem. 2000, 51, 41–63. (22) Rothenhausler, B.; Knoll, W. Nature 1988, 332, 615–617. (23) Wilkop, T.; Wang, Z. Z.; Cheng, Q. Langmuir 2004, 20, 11141–11148. (24) Wolf, L. K.; Fullenkamp, D. E.; Georgiadis, R. M. J. Am. Chem. Soc. 2005, 127, 17453–17459. (25) Shumaker-Parry, J. S.; Zareie, M. H.; Aebersold, R.; Campbell, C. T. Anal. Chem. 2004, 76, 918–929. (26) Kanda, V.; Kitov, P.; Bundle, D. R.; Mcdermott, M. T. Anal. Chem. 2005, 77, 7497–7504. (27) Brockman, J. M.; Frutos, A. G.; Corn, R. M. J. Am. Chem. Soc. 1999, 121, 8044–8051. (28) Castellana, E. T.; Cremer, P. S. Surf. Sci. Rep. 2006, 61, 429–444. (29) Phillips, K. S.; Dong, Y.; Carter, D.; Cheng, Q. Anal. Chem. 2005, 77, 2960– 2965. (30) Wang, Z. Z.; Wilkop, T.; Cheng, Q. Langmuir 2005, 21, 10292–10296. (31) Phillips, K. S.; Han, J. H.; Martinez, M.; Wang, Z. Z.; Carter, D.; Cheng, Q. Anal. Chem. 2006, 78, 596–603.

6398


approach to incorporate lipid-based molecules into a robust, airstable bilayer membrane without compromising their functionality. Several approaches have been proposed, including work by Cremer’s group which used poly(ethylene glycol) (PEG)-derivatized lipids to stabilize membranes.32,33 The reported supported bilayer has either PEG-mushroom or PEG-brush configuration, and the functionality of the air-stable bilayer membrane was largely retained. In this work, we report the development of air-stable membrane arrays for the study of the interactions of relevant proteins with phosphoinositides (PI), which are phosphorylated derivatives of phosphatidylinositol (PIPs). This class of lipids recruit and/or activate specific effectors at the cytosolic face of some intracellular membranes and has been linked to many disorders including cancer, cardiovascular disease, and immune dysfunction.34 In addition, enzymes, such as kinase, phosphatase and phospholipase, regulate membrane signal trafficking through either directly binding to PIP in the cell membrane or binding with other proteins first and the subsequent protein-protein complex binding with PIP.35 An in-depth understanding of the PIP-protein interaction would provide crucial information in identifying lipid mediators for metabolic and gene regulation and assisting biomarker discovery and drug development. Specifically, we developed addressable, lipid membrane arrays by using photolithographic methods for lipid-protein interactions with SPRi detection. A photoresist array platform was employed for this study for two principal reasons: (1) it provides ideal compartments for accommodating lipid membranes in an addressable fashion (2) and array surface modification is approachable. Several surface techniques are employed for array characterization including fluorescence, scanning electron microscopy, and electrochemistry. This flexibility in characterization and detection allows for broad-based applicability of our work. The structures of the key PIPs and other lipids used in this work are displayed in Figure 1. The five hydroxyl groups can be substituted with one, two, or three phosphate groups. As a result, there exist seven PIP isomers. For this lipid-protein interaction study, we chose to use phosphatidylinositol 4,5-biphosphate (PtdIns(4,5)P2) and phosphatidylinositol 4-phosphate (PtdIns(4)P) as the model system. PtdIns(4,5)P2 plays an essential role in cell biology and was reported to be metabolized to PtdIns(5)P during infection.36 Lowe syndrome, which is a genetic abnormality where the patient lacks the PIP-2-5 phosphatase enzyme, is also correlated with an abnormally high concentration of accumulated PtdIns(4,5)P2 in the cell.37 This syndrome causes cataracts, glaucoma, and kidney deficiencies among others. PtdIns(4)P is chosen because it is the most abundant lipid in the cell along with PtdIns(4,5)P2.34 Many PIP binding proteins are now commercially available, and the binding sites for all phosphoinositides are found to rely on the plekstrin homology (PH).37 The (32) Albertorio, F.; Diaz, A. J.; Yang, T. L.; Chapa, V. A.; Kataoka, S.; Castellana, E. T.; Cremer, P. S. Langmuir 2005, 21, 7476–7482. (33) Holden, M. A.; Jung, S. Y.; Yang, T. L.; Castellana, E. T.; Cremer, P. S. J. Am. Chem. Soc. 2004, 126, 6512–6513. (34) Rusten, T. E.; Stenmark, H. Nat. Methods 2006, 3, 251–258. (35) Kraub, M.; Haucke, V. FEBS Lett. 2007, 581, 2105–2111. (36) Niebuhr, K.; Giuriato, S.; Pedron, T.; Philpott, D. J.; Gaits, F.; Sable, J.; Sheetz, M. P.; Parsot, C.; Sansonetti, P. J.; Payrastre, B. EMBO J. 2002, 21, 5069–5078. (37) Zhang, X. L.; Hartz, P. A.; Philip, E.; Racusen, L. C.; Majerus, P. W. J. Biol. Chem. 1998, 273, 1574–1582.

Figure 1. Structure of phosphatidylinositols and other lipid molecules used in this work.

PtdIns(4,5)P2 Binding Protein (P4,5BP) contains a highly specific PH domain that recognizes and binds to PtdIns(4,5)P2.38 Ferguson et al. recently showed that the equilibrium dissociation constant (KD) of PtdIns(4,5)P2-P4,5BP (MW, 42.4 kDa) was 6.6 nM.39 At present, very little effort has been focused on new surface chemistry to reproduce a biological mimic for these lipid-protein interactions. This work aims to develop novel surface chemistry and methodology to add to the ever-expanding field of PIP-based lipidomics, by examining PIP-protein interactions with SPRi in a quantitative and high-throughput manner. We anticipate that the results obtained in a biologically mimicking environment will aid the study of lipid-protein interactions, offering useful information for understanding the metabolic pathway and gene regulation of the PtdIns(4,5)P2 lipid. EXPERIMENTAL SECTION Reagents and Apparatus. Phosphatidylcholine (PC), 1,2dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(cap biotinyl) (biotin-PE), 1,2-dioleoyl-sn-glycero-3-ethylphosphocholine (DOPC+), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE), 1,2dipalmitoyl-sn-glycero-3-phosphoethanolamine-N- (MPEG-350 PE), and L-R-phosphatidylinositol (PtdIns(4)P, PtdIns(4,5)P2) were purchased from Avanti Polar Lipids (Alabaster, AL). 2-(12-(7Nitrobenz-2-oxa-1,3-diazol-4-yl) amino) dodecanoyl-1-hexadecanoylsn-glycero-3-phosphocholine (NBD-PC) was from Molecular Probes, and monosialoganglioside receptor (GM1) was from Matreya (Pleasant Gap, PA). Phosphoinositides binding protein (PBP) and PtdIns(4,5)P2 binding protein (P4,5BP) were from Echelon Bio(38) Lemmon, M. A. Traffic 2003, 4, 201–213. (39) Ferguson, C. G.; James, R. D.; Bigman, C. S.; Shepard, D. A.; Abdiche, Y.; Katsamba, P. S.; Myszka, D. G.; Prestwich, G. D. Bioconjugate Chem. 2005, 16, 1475–1483.

sciences (Salt Lake City, UT). ImmunoPure Avidin was from Pierce Biotechnology. Fatty-acid free bovine serum albumin (FFBSA), cholera toxin (CT), anti-CT IgG, 3-mercapto-1-propanol (MPA), 3-(mercaptopropyl) trimethoxysilane, and potassium ferricyanide were purchased from Sigma-Aldrich. Potassium ferrocyanide was from Fisher Scientific. Other chemicals are all analytical grade. All buffer solutions were prepared with 18 MΩ Milli-Q water, filtered with a Nalgene 0.2 µm filter, and degassed for 15 min before use. Surface plasmon resonance imaging was performed with a built-in-house SPR imager, as described in previous work.23 A Biosuplar II instrument (Analytical µ-Systems, Germany) was used for SPR spectroscopy measurements. Surface characterization of the photopatterned photoresist array and supported bilayer array was conducted with a Leica Fluo MZ III stereomicroscope, a Philips XL30 FEG scanning electron microscope (SEM), and a CHI 650 electrochemical workstation. Preparation of Vesicles. PC (or PE, DOPC+), PE-PE, and certain amounts of PtdIns(4)P or PtdIns(4,5)P2 were mixed in chloroform and dried with a stream of nitrogen to form a dry lipid layer in a glass vial. After rehydration with Tris buffer (10 mM, pH 7.5, 0.15 M NaCl), the suspension was sonicated with a sonifier (model 250, Branson Ultrasonics) in an ice bath for 30 min. The vesicle solutions were then centrifuged at 14 000 rpm for 15 min. The clear vesicle solutions were transferred to a clean cryogenic vial, followed by 10 freeze-thaw cycles. Finally, the vesicles were extruded through a polycarbonate filter with pore size of 100 nm. The vesicles were then stored in a 4 °C refrigerator and ready for use within 3 days. Preparation of Photoresist Arrays. SPR gold substrate was prepared as described in previous publications.23 Microposit S1813 positive photoresist (Rohm and Haas, Marlborough, MA) was spincoated onto an SPR gold substrate at 6000 rpm for 30 s with a Analytical Chemistry, Vol. 80, No. 16, August 15, 2008

6399

Bidtec SP100 spin-coater. The thickness of the resulting photoresist layer is about 2 µm. This substrate was then soft-baked at 60 °C for 1 min and then covered by aluminum foil to avoid any photodegradation. After cooling down to room temperature, a photomask was placed on top of the photoresist layer and was exposed to a UV lamp at 10 mM/cm2 for 14 s using an OAI light source system (OAI, Milpitas, CA). The photoresist was then developed in a M351 (Rohm and Haas, Marlborough, MA) developer (1:5 v/v in DI water) for 1 min, with gentle swirling of the developer. The remaining photoresist was then hard-baked for 2 min at 60 °C. Cyclic Voltammetry and Impedance Measurements. A CHI 650 electrochemical workstation (CH Instruments, Austin, TX) was employed in both measurements. In both cases, a threeelectrode system was used, with a 2 mm gold electrode (or a gold substrate defined by a 2 mm in diameter O-ring) as the working electrode, platinum wire as the counter electrode, and Ag/AgCl as the reference electrode. The working solution consists of 1 mM Fe(CN)64-/3-, with 0.1 M KCl as a supporting electrolyte. Cyclic voltammetry (CV) was conducted at a scan rate of 50 mV/ s. The applied potential for impedance measurements was 0.191 V, based on the cyclic voltammetry results. The data fitting was performed with CHI software. Array Characterizations by Fluorescence Microscopy. Fluorescent vesicles were prepared by doping 3 molar % NBDPC into the vesicles. Pristine images were taken due to the autofluorescence nature of the photoresists with an emission wavelength of 650 nm. Green fluorescence images were obtained at 534 nm, after the fluorescent vesicles fused on the patterned substrate. Stability Tests of Bilayer Arrays. 3-Mercapto-1-propanol (MPA) was self-assembled on the gold substrate for 12 h. The vesicles were then injected into a SPR flow cell, and then the flow was stopped to allow complete vesicle fusion on the hydrophilic MPA modified surface. The dehydration of the supported lipid bilayer was conducted by injecting an air bubble into the SPR cell and keeping it in a dry state for at least 30 min. Finally, the running buffer was resumed to rehydrate the bilayer membrane. Three SPR spectra, pristine MPA, SLM, and rehydrated SLM, were taken, respectively. SLM Biosensing with SPR Imager. The patterned photoresist array was immersed in 1 mM MPA for 12 h. Vesicles of different compositions were then manually pipetted onto each array element, followed by incubation for 1 h in a humidity chamber. After the lipid array was carefully washed with Tris buffer, it was incubated with 3% (w/v) fatty acid-free BSA for 1 h. This chip was then washed with Tris buffer three times and ready for SPR imaging analysis. Pristine SPR images and those after incubation with target protein analytes were sequentially obtained in air. The resulting difference images were employed for further analysis. RESULTS AND DISCUSSION Array Fabrication and Characterization. The SPR image of an array patterned substrate obtained via photolithography is shown in Figure 2. Microposit S1813 positive photoresist was used for array fabrication, and the photomask was generated by using a high resolution printer on transparency. The chip has four groups of identical elements, and each element consists of 3 × 3 6400


Figure 2. An SPR image of the photoresist arrays for lipid membranes fabricated with the photolithographic method.

wells, each with a dimension of 300 × 300 µm, with a 300 µm distance between each array element. The array pattern has also been characterized by fluorescence microscopy, which shows that the photolithography technique is capable of generating welldefined patterns, providing an ideal platform for further vesicle fusion and confinement of lipid bilayers within each array element. These arrays could be reproduced with very small coefficients of variation in element size and shape. Previous results indicate the cleanliness of the gold substrate affect the formation of the self-assembled monolayer significantly. In order to determine the efficiency of photoresist removal, a 100mesh TEM grid was used as a photomask, and SEM confirmed an acceptable level of cleanliness of the photoresist-removed gold surface (data not shown). It must be noted that despite the relatively clean image in SEM, photoresist debris may still remain on the surface at the submicrometer level or even smaller scales. It is important to confirm that these photoresist residuals or “islands” would not affect the formation of SAMs, supported lipid bilayer, or receptor functionality. A series of experiments of lipid formation on such a surface were conducted and will be discussed later in this report. Bilayer Formation and Stability Test. An important issue with regards to the lipid membrane is whether the lipid bilayer is formed inside the patterned wells and if the bilayers are stable after undergoing dehydration and rehydration procedures that are typically required during the fabrication of addressable lipid bilayer arrays. To verify the bilayer formation, both direct and indirect characterization methods have been implemented. SPR spectroscopy is known to be effective for monitoring the vesicle fusion and characterizing the lipid membrane thickness through simulation and data fitting. In our experiment with the Biosuplar II instrument, a ∼0.5° angular shift was typically found for vesicle fusion on a pristine MPA modified gold surface. On a substrate treated with photoresist cast and removal followed by MPA

Figure 3. Cyclic voltammograms of photoresist-removing gold modified with SAM (solid line) and after vesicle fusion (dashed line) in 1 mM Fe(CN)64-/3- solution with 0.1 M KCl as the supporting electrolyte. The scan rate is 50 mV/s.

modification, the resonance angular shifts fell in a range between 0.5 and 0.8°, largely depending on the composition and the nature of the lipids. Considering that there might be trace amounts of photoresist debris remaining on the surface, one can conclude that the results reasonably indicate the same fusion process on these surfaces. To further verify the vesicle fusion measured by the SPR method, cyclic voltammetry was performed using modified gold as the working electrode. The cyclic voltammogram shown in Figure 3 displays a well defined response for Fe(CN)4-/3-, indicative of a mass transfer-controlled process on a SAM modified surface, likely due to the short chain of MPA. Upon vesicle fusion, the lipid bilayer blocked the free access of the redox probe to the surface, resulting in a “flat” charging current, which strongly suggests the formation of a lipid layer on the hydrophilic SAM surface. It should be noted that the charging current is not totally flat, especially at high potential, indicative of a “leaky” membrane. This is quite common for supported bilayer membranes, likely due to the existence of defects. An ac impedance measurement was also conducted to determine the bilayer thickness and verify the SPR and CV results. Figure 4a shows the impedance results of gold with SAM assembled on top of it. Figure 4b shows impedance results of the lipid layer on top of the SAM layer after fusion. A simple equivalent circuit, as shown in Figure 4b, was used to perform data fitting. RCT was found to be 10 kΩ (vs 382 Ω for Au/MPA), which suggests the lipid membrane effectively blocked the electron transfer. Furthermore, the thickness of the lipid membrane was calculated by using eq 1:

C)

Aε0ε d

(1)

where C represents the capacitance of a parallel plate capacitor, A is the area of working electrode, ε is the effective dielectric constant of the lipid membrane, and ε0 is the permittivity of the

free space. Taking an ε value of 2.7,40 the calculated thickness of the lipid layer is 4.7 nm, which agrees well with the literature value of a bilayer membrane and thus confirms the SPR measurement described above. An indirect characterization method was also employed to prove the lipid bilayer formation. This method takes advantage of the fluorescence quenching effect due to energy transfer when a fluorescence dye is proximate to a metal surface,41 which happens in this case. When the photoresist array template was incubated with vesicle solutions, two distinct surface areas on the array underwent different physical changes. Autofluorescence of the photoresist enables a red fluorescence image to be taken at a wavelength of 650 nm, as shown in Figure 5a. When fluorescent NBD-PC doped vesicles were fused onto the array template, the fluorescence image displays a strong green fluorescence in the photoresist area and dark areas in the photopatterned wells (Figure 5b). On the hydrophobic photoresist surface, the formation of a hybrid bilayer appears to take place.42,43 The bright bulk fluorescence area is due to the fact that the photoresist layer has a thickness larger than 100 nm, lifting the fluorescence dye beyond the quenching distance. On the other hand, inside the photoresist array elements (with SAM assembled), there are two possible processes for the vesicles: (1) intact vesicles packing closely inside the wells and (2) bilayer formation through vesicle fusion on the hydrophilic SAM surface. If vesicles retain their structure and pack tightly inside the well, a considerable degree of fluorescence signal should be observed as the size of the vesicle (∼100 nm) will elevate some of the fluorescence dyes beyond the quenching range.30 The darkness inside the wells firmly indicates the formation of a planar lipid layer as the fluorescence signal has been substantially quenched by the gold and completely rules out the possibility of the existence of any intact vesicles. From SPR spectroscopy, CV, ac impedance, and fluorescence measurements, all results point to successful lipid layer formation on the hydrophilic SAM surface inside the array elements. The next important question is whether the lipid bilayer membrane can be air stable. Stability is essential to the construction of an addressable membrane array because a dehydration procedure is usually encountered. Albertorio et al. has reported that a lipid bilayer can be stabilized by 7 molar % poly(ethylene glycol) phosphatidylethanolamine (PEG-PE).32 Here we tested three kinds of lipid bilayers containing PEG-PE with SPR spectroscopy as described in the Experimental Section. The recovery of the SPR angle after a dehydration-rehydration process was found to be correlated to membrane integrity and thus used to assess the air stability of the film. For instance, fusion by the DOPC+/PEG-PE vesicle on the MPA surface led to a 0.781° angular shift. After the lipid membrane underwent dehydration and rehydration, the resonance angle shifted toward smaller angle by 0.018°, which produces a recovery efficiency of 98%. With the use of this approach, the bilayer recovery for PE/PEG-PE was found to be 97%. However, the PC/PEG-PE bilayer membrane is subject to considerable damage during the process, only showing (40) Plant, A. L.; Gueguetchkeri, M.; Yap, W. Biophys. J. 1994, 67, 1126–1133. (41) Lakowicz, J. R. Anal. Biochem. 2001, 298, 1–24. (42) Duschl, C.; Liley, M.; Corradin, G.; Vogel, H. Biophys. J. 1994, 67, 1229– 1237. (43) Jenkins, A. T. A.; Neumann, T.; Offenhausser, A. Langmuir 2001, 17, 265– 267.


6401

Figure 4. Alternating current impedance spectroscopy of (a) the photoresist-removing gold modified with SAM and (b) after vesicle fusion in 1 mM Fe(CN)64-/3- solution with 100 mM KCl as the supporting electrolyte. The insert is the depiction of the equivalent circuit model.

Figure 5. Fluorescence images of (a) the photoresist array and (b) after fusion of NBD-PC doped vesicles on the same substrate.

a 68% recovery, which differs from the result of Albertorio et al.32 The reason for this is still unclear; and it could be related to the difference in the preparation steps. We therefore chose the DOPC+/PEG-PE membrane as the biosensing platform throughout the rest of our experiments. To investigate the effect of photoresist debris on bilayer formation, these experiments were also performed using DOPC+/PEG-PE on a photoresist-removed surface. The results were similar to those on the clean Au surface, showing the bilayer recovery was also near complete on this surface (data not shown). This suggests that the photoresist residual, if there is any, does not affect the lipid layers. Biosensing with the Air-Stable Lipid Membranes. We next characterized if the rehydrated lipid membranes retain their biological functionality, especially after addition of 7 molar % PEGPE. For this test, 5 molar % biotin-PE, GM1, or PtdIns(4,5)P2 was doped into the vesicles as capturing ligands to detect avidin, cholera toxin (CT), and PtdIns(4,5)P2 binding proteins, respectively. The lipid membranes on the MPA surface were dehydrated and rehydrated in an SPR flow cell, and the binding response was measured following the injection of analytes. Three calibration curves were obtained with excellent linearity, suggesting good 6402


functionality of the bilayer membranes. The sensitivity and correlation coefficient values for avidin (slope ) 0.0177°/µg/mL; R2 ) 0.9834), cholera toxin (slope ) 0.0149°/µg/mL; R2 ) 0.9829), and PtdIns(4,5)P2 binding protein (slope ) 0.0489°/µg/mL; R2 ) 0.9826) were all very good. The limit of detection was determined to be 33.3 nM for avidin, 73.5 nM for cholera toxin, and 25 nM for PtdIns(4,5)P2 binding protein, which are all comparable to literature values.39 These results are significant because they are obtained on rehydrated lipid bilayers, indicating these films are air stable and thus suitable for the planned array fabrication/detection in ambient. A control experiment was conducted with the GM1/CT model system to compare the sensitivity on a pristine lipid membrane and that of a dehydrated/ rehydrated membrane for detection of 50 µg/mL CT. No significant difference was observed in their SPR signals (data not shown), suggesting the dehydration/rehydration process does not seem to affect the functionality of these membranes and, most importantly, the sensitivity. An additional control experiment was performed with BSA on the dehydrated/rehydrated membrane containing GM1. Results indicated that the nonspecific interactions on the rehydrated membranes were insignificant.

Figure 6. Screening of bioaffinity interactions using lipid membrane arrays with SPR imaging. (a) Interactions of 0.05 mg/mL avidin with lipid membranes of PtdIns(4)P/DOPC+/PEG-PE (1), biotin-PE/DOPC+/PEG-PE (2), DOPC+/PEG-PE (3), and GM1/DOPC+/PEG-PE (4). (b) Interactions of 0.05 mg/mL cholera toxin (CT) with lipid membranes of GM1/DOPC+/PEG-PE (1), biotin-PE/DOPC+/PEG-PE (2), DOPC+/ PEG-PE (3), and PtdIns(4)P/DOPC+/PEG-PE (4). (c) Interactions of 0.002 mg/mL PtdIns(4,5)P2 binding protein with lipid membranes of PtdIns(4,5)P2/DOPC+/PEG-PE (1), biotin-PE/DOPC+/PEG-PE (2), PtdIns(4)P/DOPC+/PEG-PE (3), and DOPC+/PEG-PE (4). (d) A binding profile of the biotin-avidin interaction shown in part a.

SPR Imaging Analysis of PIP/PIP-Binding-Protein Interactions. An important goal of this work is to test the feasibility of fabricating lipid arrays in ambient and carry out the PIP interaction analysis in a high-throughput fashion with SPR imaging. With the use of the information obtained previously on the air-stable supported membranes, five types of lipid vesicles, DOPC+/PEG-PE, GM1/DOPC+/PEG-PE, biotin-PE/DOPC+/ PEG-PE,PtdIns(4)P/DOPC+/PEG-PE,andPtdIns(4,5)P2/DOPC+/ PEG-PE, were employed for the construction of the addressable bilayer arrays. Biotin-avidin and GM1-CT were used as model systems because they have been extensively studied.31 Figure 6 shows the SPR difference images for several systems studied here. For a system containing PtdIns(4)P/DOPC+/PEG-PE, biotin-PE/ DOPC+/PEG-PE, DOPC+/PEG-PE, and GM1/DOPC+/PEG-PE, incubation with 0.05 mg/mL avidin shows the strongest binding occurring between avidin and biotin-PE/DOPC+/PEG-PE bilayers (Figure 6a). Little signal is observed for PtdIns(4)P/DOPC+/ PEG-PE and DOPC+/PEG-PE lipid membranes. However, crossreactivity between avidin and GM1/DOPC+/PEG-PE bilayer was

clearly observed, likely due to the five carbohydrate units on the GM1 head groups. The line profile of the biotin-avidin interaction is displayed in Figure 6d, showing a satisfactory homogeneity inside the well and between the wells. The minimal background signal for DOPC+/PEG-PE layer indicates the nonspecific interaction is insignificant. The cholera toxin-GM1 interaction was also investigated with the lipid membrane arrays. Four lipid membranes, GM1/DOPC+/ PEG-PE, biotin-PE/DOPC+/PEG-PE, DOPC+/PEG-PE, and PtdIns(4)P/DOPC+/PEG-PE were employed for constructing the array elements. As expected, CT-GM1 binding was the strongest among all the ligands we employed (Figure 6b). Interestingly, there is no visible cross-reaction between CT and PtdIns(4)P or biotin-PE, and the nonspecific interaction on DOPC+ is insignificant. To investigate the interactions involving PtdIns(4,5)P2 binding protein, the lipid membrane array was fabricated with PtdIns(4,5)P2/DOPC+/PEG-PE, biotin-PE/DOPC+/PEG-PE, PtdIns(4)P/DOPC+/PEG-PE, and DOPC+/PEG-PE. The SPR Analytical Chemistry, Vol. 80, No. 16, August 15, 2008

6403

imaging result is shown in Figure 6c. Clearly the PtdIns(4,5)P2/ DOPC+/PEG-PE arrays show the strongest binding signal while the signals from the DOPC+/PEG-PE and biotin/DOPC+/PEGPE bilayers are barely visible. A slight degree of cross-reaction with the PtdIns(4)P/DOPC+/PEG-PE bilayer was observed, but the binding was considerably weaker, which is further verified with the SPR spectroscopy measurements. Two control experiments were conducted with SPR spectroscopy to confirm the binding selectivity. First, the PtdIns(4,5)P2/DOPC+/PEG-PE membrane was incubated with 2 µg/mL FFBSA and no response was observed. Second, 2.5 µg/mL PtdIns(4)P binding protein (PBP) was incubated with the PtdIns(4,5)P2/DOPC+/PEG-PE bilayer, the binding response was only 30% as compared to the signal from P4,5BP to PtdIns(4,5)P2. These experiments provide additional evidence for profiling the specific interactions involving PtdIns(4,5)P2, further verifying the results from the array approach. Compared to traditional methods, this new, air-stable lipid membrane array, when coupled with the SPR imaging technique, shows tremendous advantages for its straightforward and nonlabeling detection capability.

thickness values, obtained from SPR and impedance, seem to agree well with the theoretical value for a planar morphology. The supported phospholipid bilayers show a varied degree of robustness while undergoing dehydration and rehydration processes, dependent on the composition. The DOPC+/PEG-PE layers exhibit an amazing 98% recovery after the dehydration/rehydration process, while the nonspecific interaction on this surface remains quite low. These films also demonstrate a highly preserved functionality for sensing applications. The detection sensitivity, as compared to that on a pristine membrane, is not affected by the rehydration process. The use of uniquely constructed lipid array for screening of lipid-protein interactions with SPR imaging has been demonstrated with PIP molecules. Multiplexed detection of several lipid-protein systems was examined in a quantitative and high-throughput manner, which shows high potential for the system to be used for large scale proteomic analysis. The results clearly proved that SPRi is a promising tool for array-type analysis in biomimetic membrane-based studies. Future work will focus on the screening of these protein interactions in a real-time, in situ manner.

CONCLUSIONS We have demonstrated a simple and effective method to fabricate lipid membrane arrays by photolithography. A series of surface characterization methods, including fluorescence, electrochemistry, SPRi, and scanning electron microscopy, have been employed to validate the array quality and the bilayer formation. The resulting array surfaces appear to be well-defined and homogeneous. The formation of lipid bilayers was first characterized using cyclic voltammetry and ac impedance measurements directly and then by fluorescence microscopy indirectly. The layer

ACKNOWLEDGMENT We would like to thank Dr. Pingyun Feng and Dr. Yang Liu for discussion of ac impedance measurements. This work was supported by the National Science Foundation (Grant CHE0719224).

6404


Received for review April 25, 2008. Accepted June 11, 2008. AC800845W

Development of Air-Stable, Supported Membrane Arrays with

Recommend Documents